ChanServ changed the topic of #glasgow to: glasgow debug tool · code https://github.com/GlasgowEmbedded/glasgow · forum https://glasgow.whitequark.org · logs https://freenode.irclog.whitequark.org/glasgow · production https://www.crowdsupply.com/1bitsquared/glasgow · no ETAs at the moment
Stormwind_mobile has joined #glasgow
oeuf has quit [Ping timeout: 260 seconds]
egg has joined #glasgow
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
Exec1N has joined #glasgow
<_whitenotifier-f> [libfx2] artizirk commented on pull request #4: Add standalone url field to fx2 setup.py - https://git.io/JfPR7
ExeciN has quit [Disconnected by services]
Exec1N is now known as ExeciN
<_whitenotifier-f> [libfx2] artizirk commented on pull request #4: Add standalone url field to fx2 setup.py - https://git.io/JfP0j
<d1b2> <tnt> Anyone know what the clkref frequency is from memory ?
<d1b2> <edbordin> crystal is 24 MHz if that's what you're after
<d1b2> <tnt> No, the CLKREF pin from the FX2 to the FPGA.
<d1b2> <edbordin> nvm, back to lurking for me
futarisIRCcloud has joined #glasgow
<d1b2> <gimbas> just thought of a possible applet, cause i needed it just now xd pulse generator, pwm generator
<d1b2> <gimbas> maybe it exists already and i don't know about it?
<d1b2> <gimbas> either way my glasgow isn't even assembled yet xd
<whitequark> tnt: 48 MHz
<tnt> whitequark: tx. Unfortunately I've confirmed the HX PLL are different from the UP5k ones. Previous gen I guess.
<whitequark> oh
<whitequark> do you still only have revB?
<tnt> No no, I have both revB and revC glasgow. But I was testing the divide-by-5 mode of the PLL for TMDS 10:1 but that only works on the UP5k, the HX doesn't have that.
<whitequark> huh
Getorix has joined #glasgow
Getorix_ has quit [Ping timeout: 256 seconds]
Getorix_ has joined #glasgow
Getorix has quit [Ping timeout: 260 seconds]
Stormwind_mobile has quit [Ping timeout: 265 seconds]
Stormwind_mobile has joined #glasgow
tomtastic_ has joined #glasgow
tomtastic has quit [Ping timeout: 246 seconds]
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
englishm has quit [Excess Flood]
englishm has joined #glasgow
tomtastic_ is now known as tomtastic
rappet has quit [Quit: -]
rappet has joined #glasgow
bvernoux has joined #glasgow
ExeciN has quit [Quit: so long king Bowser]
Stormwind_mobile has quit [Ping timeout: 246 seconds]
Stormwind_mobile has joined #glasgow
<electronic_eel> esden: whitequark: what do you think about changing the adc to INA233 for revC2?
<whitequark> didn't we discuss it before?
<whitequark> I *think* I agreed to it?
<electronic_eel> yes, and as niklas was doing the edinburgh design, we wanted to test it there
<whitequark> ahh
<electronic_eel> but it seems this design was never finished, so we didn't get results
<whitequark> mh ok
<whitequark> also now I remember that I owe you some I2C stuff...
<electronic_eel> so now I think the question is if esden wants to do it
<whitequark> been working mostly on nmigen in the meantime since that's what puts food on my table, quite literally
<electronic_eel> yeah, the interface to the internal i2c
<whitequark> I should find some time to implement that + an ELA
<electronic_eel> that will be needed when we want to make the test jig board work
<electronic_eel> wasn't someone else working on the ELA?
<electronic_eel> awygle?
<whitequark> I suggested that I do it since I wrote all of the existing code for --trace and it should be very easy for me to adapt it
<awygle> whitequark told me to stop :-P
<whitequark> it's probably like a hour of work
<whitequark> well, it's not that I wanted you to stop
<awygle> I am joking
<whitequark> right ok
<awygle> I do still want to write a Glasgow applet, and ideally document the process, but if wq is willing to do the ELA I have other higher value things to do in the short term
<electronic_eel> whitequark: another question - back when we talked about the self test (which still has to be fixed for revC), I said that we'll need the internal pullups of the ice40
<whitequark> the problem with documenting the process is that the process is bad
<whitequark> *everything* about applets is in flux
<electronic_eel> and you said that nmigen currently has no way of accessing those
<electronic_eel> is this still the case?
<whitequark> they're halfway migrated to nmigen, they're halfway using nmigen's platform layer...
<whitequark> electronic_eel: uhm, not quite
<electronic_eel> maybe I remember it wrong
<whitequark> it's not an nmigen problem
<whitequark> IIRC it's a glasgow-using-nmigen problem
<whitequark> I don't recall the discussion, think you can find it in the logs?
<whitequark> or let's just rehash it
<electronic_eel> oh, so nmigen can use them, but glasgow is still using the omigen compat layer?
<whitequark> no
<whitequark> there's no compat layer for the platform stuff
<whitequark> you can add PULLUP=1 in rev_c0.py
<electronic_eel> this is the old discussion
<whitequark> oh
<whitequark> let's see
<whitequark> I may have been just wrong there
<whitequark> so let me explain how this all works
<whitequark> on ice40, you can set IOSTANDARD, PULLUP, and so on as parameters on SB_IO
<whitequark> in nmigen, you normally provide these in the platform, in the resources list
<whitequark> unfortunately this doesn't really cover meta use cases like glasgow, which add them at runtime
<whitequark> or rather, the way glasgow uses resources isn't appropriate for its meta use case
<electronic_eel> runtime as in python running nmigen elaborate or runtime as in logic switching in the fpga?
<whitequark> neither
<whitequark> nmigen.build is mildly biased towards fully static FPGA pinouts since most boards work that way
<whitequark> it provides a connector functionality for PMODs etc
<whitequark> glasgow right now pretends to be a fully static board with two 8-pin IO resources
<whitequark> this isn't how applets actually work and it hurts glasgow in more ways than one
<whitequark> what *ought* to happen is that there would be no resources defined upfront, only connectors corresponding to ports
<whitequark> (in fact we can literally rename the FPGA pins to A0..A7 B0..B7)
<whitequark> and then each applet would create a resource with the pins it needs
<whitequark> so it'd be like on-demand construction of board definitions
<whitequark> in that case it's very easy to say if we want pullups, or DDR, or whatever
<whitequark> right now glasgow has it hacked together in a fairly obtuse way
rappet has quit [Ping timeout: 272 seconds]
<whitequark> anyway
<whitequark> forget all that
<whitequark> should we maybe just enable the pullups permanently on every design?
<whitequark> it's usually not very good to have floating digital inputs
<whitequark> then that would be a 2-line patch for the existing code
rappet has joined #glasgow
<electronic_eel> so for the selftest, we would have stuff like pullups & input everywhere and on pin is output low, read back all input pin status
<electronic_eel> that would work with pullups enabled always
<whitequark> shall I enable the inputs now?
<electronic_eel> inputs? do you mean pullups?
<whitequark> er, yes
<electronic_eel> if it is just a 2-line patch, maybe put it in a branch and let me play a bit with it?
<whitequark> is there a reason we shouldn't just have pullups on ice40 side always enabeld
<whitequark> ?
<electronic_eel> what happens with unused pins that go to the levelshifters currently?
<electronic_eel> and especially the dir lines to the level shifters?
<electronic_eel> I think the dir should be set to low, to set the level shifter to input
<electronic_eel> then the pin to the ice40 is driven by the level shifter
balrog has quit [Ping timeout: 260 seconds]
<whitequark> electronic_eel: right now all unused pins are tied low
<tnt> That's new ? Because it was the case and got "fixed" but that was back in rev B.
<electronic_eel> ok, so they are actively driven low. additional pullups would just be some µa flowing
<whitequark> sorry, I misspoke
<tnt> It's annoying that unused pins are driven (or even pulled) because if you're switching from one applet to another using different pins, you're messing with the DUT.
<whitequark> unused OEs are driven low
<whitequark> so the pins are hiz
<tnt> ok
<electronic_eel> yes, that is how I think it makes sense
<whitequark> see target/hardware.py
<electronic_eel> so then I don't see an issue with always enabling the pullups on the pins to the level shifters
<electronic_eel> we may not enable pullups on the sync though
<whitequark> right
<electronic_eel> haven't thought about the pins to the lvds port, there pullups would probably make sense
<whitequark> wait, I think we can enable the PU on revC1's sync
<electronic_eel> you mean because the pulldown resistor is strong enough?
<whitequark> revC1's sync has an LVC1T45 on it, right?
<electronic_eel> yes
<whitequark> so on ice40's side we can enable the pullup on io while oe isn't enabled, right?
<whitequark> or am I forgetting something
<electronic_eel> on the io yes, the dir pin is the interesting part
<electronic_eel> there is R10, a 1K pulldown on the dir line
<whitequark> I wasn't thinking about enabling pullups on DIR pin
<electronic_eel> ok
<whitequark> is there any case where it's useful?
<electronic_eel> selftest
<whitequark> shorts?
<electronic_eel> yeah, but for the sync it is hard, because we have nothing else internally that can read the status back
<electronic_eel> for the regular pins we have the pullup/pulldown-i2c-expanders
<whitequark> the problem with pullups on OE pins is that currently that won't do anything
<electronic_eel> but the test jig board has a line for sync
<whitequark> because their direction is strictly output
<electronic_eel> hmm, there is no currently no gateware for syncing to another glasgow?
<whitequark> nop
<whitequark> and i don't even really know how to do it reliably
<electronic_eel> I thought I remember you posting something with 2 glasgows or something?
<whitequark> i don't think so
<electronic_eel> or do I misremeber?
<electronic_eel> maybe some dirty hack
<whitequark> hm
<whitequark> i don't remember making one, but then my memory is pretty holey
<electronic_eel> for testing purposes we'd just need a simple input, I think that shouldn't be too hard
balrog has joined #glasgow
balrog has quit [Ping timeout: 246 seconds]
<_whitenotifier-f> [GlasgowEmbedded/glasgow] whitequark pushed 1 commit to master [+0/-0/±2] https://git.io/JfPhj
<whitequark> okay, now selftest always fails with:
<whitequark> E: g.applet.internal.selftest: pins-int: fail high: A0 A1 A2 A3 A4 A5 A6 A7 B0 B1 B2 B3 B4 B5 B6 B7
<_whitenotifier-f> [GlasgowEmbedded/glasgow] whitequark 8c4d624 - platform.rev_c[01]: add pullups on IO pins.
<electronic_eel> I think that is to be expected, because they are expected to have keepers on them
<whitequark> yep
<whitequark> that was in the sense of "the patch works reliably"
<electronic_eel> yeah, better reliably known fail than have this floating mess
<electronic_eel> ok, so one issue down, especially one where I thought it might take you some time to fix the nmigen design
<electronic_eel> very good progress
<whitequark> can someone remind me again why do we have a forum?
<whitequark> I have no recollection of why it exsits
<whitequark> and there were only two posts in all that time, indicating that it's maybe not that useful
<electronic_eel> I don't know, I think you wanted to have one back some months
<whitequark> esden: do you remember why it exists, perhaps?
<electronic_eel> I think the discord<->irc bridge is better, combined with the search function in your irclog
<whitequark> yeah I'm leaning towards getting rid of it
<electronic_eel> and anyone having a problem can also still open a github issue
ChanServ changed the topic of #glasgow to: glasgow debug tool · code https://github.com/GlasgowEmbedded/glasgow · logs https://freenode.irclog.whitequark.org/glasgow · discord https://1bitsquared.com/pages/chat · production https://www.crowdsupply.com/1bitsquared/glasgow · no ETAs at the moment
<_whitenotifier-f> [GlasgowEmbedded/glasgow] whitequark pushed 1 commit to master [+0/-0/±1] https://git.io/JfPjB
<_whitenotifier-f> [GlasgowEmbedded/glasgow] whitequark da6aa8a - README: remove forum link.
<electronic_eel> spring cleaning ;)
<d1b2> <esden> I don't know when the the forum got created.
<d1b2> <esden> It really depends on the user type which tool they prefer to get support through.
<d1b2> <esden> but having discord for icebreaker indeed makes the forum very sparsely frequented
<whitequark> the main problem with the current forum script is that it doesn't email me on a new topic and can't be made to
<whitequark> and... i would prefer people to go to github issues over fussing with a different script
<d1b2> <esden> discourse forum that I installed works very well from notification and ease of administration perspective
<whitequark> i absolutely am not going to maintain a discourse installation
<d1b2> <esden> I am actually thinking of expanding it to be an overarching forum for 1bitsquared products/projects in general
<whitequark> i took one look at the requirements and it's an IT nightmare
<whitequark> i guess we could use 1b2's discourse instance
<electronic_eel> that would be another idea, if there is really demand for it
<d1b2> <esden> right I just installed their docker thing... and it "maintains itself"
<d1b2> <esden> I just need to press the "update button" from time to time 🙂
<electronic_eel> hmm, I have my doubts about this kind of maintaining
<d1b2> <esden> @whitequark sure you are welcome to use my instance...
<whitequark> let's say i'm not very happy about it either
<ktemkin> there are some uniquely wonderful things about forums, too -- where else can you find a google result of yourself asking the same question five years earlier?
<d1b2> <esden> I need to move the icebreaker forum and expand it
<whitequark> but as long as the forum is someone else's problem i'm ok having it
<d1b2> <esden> it feels wasteful to use the forum just for one project somehow
<whitequark> ktemkin: ha!
<d1b2> <esden> there is not enough traffic to justify multiple forums
<d1b2> <esden> @Kate Temkin hahaha! 😄
<d1b2> <esden> INDEED....
<whitequark> i'm somewhat surprised how well the discord bridge works
<whitequark> but that's all good; i think i'm a lost cause in that i can't reliably stay logged into discord
<ktemkin> esden: you stole qyriad's setup, right?
<electronic_eel> @esden does atting highlight my message in your discord?
<d1b2> <esden> @Kate Temkin I even asked @Qyriad for advise and she gave me lot's of good scripts and pointers. 😄
<ktemkin> if so, if you want, I can just ask her to put your discord bridges on our server
<d1b2> <esden> eloctronic_eel it does ... but because you don't have an account here @ ing you is bit annoying. 😄
<whitequark> yeah, not having tab-completion is the most annoying part of the bridge
<whitequark> and it's not that annoying
<whitequark> matrix does work better in this aspect
<ktemkin> there was a nicer-looking project I considered when looking at bridges originally
<d1b2> <esden> @Kate Temkin I almost asked @Qyriad but I fealt that it would be rude to put additional admin burden on her. ... right <@!199166349715570688> ? it would be rude ... 😄
<ktemkin> it ran an IRC _server_ for the discord instead of a bridge
<whitequark> "... right <@!199166349715570688> ?" heh
<electronic_eel> matrix works quite nicely, we are using it at work. but I have some usability grief with the riot webclient.
<whitequark> wait, let me see how it actually works discord-side
<d1b2> <Qyriad> @esden if you already have the configuration…configured, then it's almost no work at all
<d1b2> <Kate Temkin> I think the configuration is the same as ours
<d1b2> <Qyriad> Well, just about
<d1b2> <Qyriad> Well, just about
<d1b2> <esden> @Qyriad that is true. But setting it up just took about a day of cussing, so not that bad. I decided to put it on it's own linode server. So in the future I can put other bridge like things on there too.
<d1b2> <esden> I did not want to contaminate my dedicated rented server with the node stuff...
<d1b2> <Qyriad> lol
<d1b2> <whitequark> @esden
<d1b2> <whitequark> it roundtrips but looks weird on irc 😄
<d1b2> <Qyriad> Huh
<d1b2> <esden> Oh ok the bridge did a bubu there
<d1b2> <Qyriad> That's surprising
<electronic_eel> oh, a whitequark from the dark side ;)
<d1b2> <whitequark> lol
<d1b2> <esden> so you can't at a person twice in a line ... interesting
<d1b2> <Kate Temkin> I wonder if it only substitutes the IDs the first time
<d1b2> <Qyriad> @Qyriad
<d1b2> <Kate Temkin> a @Qyriad is as a <@!199166349715570688> does
<ktemkin> that's gotta be a bug
<d1b2> <whitequark> someone forgot //g? 😄
<d1b2> <Qyriad> Probably
<d1b2> <esden> Ok folks... I need to run... (I need to do a dump run with our trash and my inlaws trash... ... of all things)
<d1b2> <esden> If the double @ bug is fixed I can updated the bridge... I probably should set a reminder to update the bridge from time to time...
<ktemkin> Qyriad: want to fix that upstream for me? >.>
<d1b2> <Qyriad> I'll take a look
<ktemkin> thanks
<ktemkin> make sure you charge it as work time =P
<d1b2> <esden> @Qyriad you are too nice ❣️ 🙂
<d1b2> <esden> out... ttyl
<d1b2> <Qyriad> Have fun~
<hl> whitequark: just received word of this massive dump of leaked rockchip datasheets: https://mega.nz/folder/2w0gQACZ#h7cbYShN_S95ZE-JRvVD3A mirror away
balrog has joined #glasgow
<whitequark> need an app to download :/
<d1b2> <gimbas> hm, whi is rockchip?
<d1b2> <gimbas> do edits make it to the IRC? i imagine not
<whitequark> ah the datasheets are ok
<whitequark> gimbas: nope they do not
<hl> whitequark: install the OSS CLI app 'megadl' (package 'megatools'), run megadl https://mega.nz/#f!2w0gQACZ!h7cbYShN_S95ZE-JRvVD3A in an empty directory
<hl> (megadl doesn't understand new-style URLs so you have to reformulate them to the above format)
<whitequark> ah
bvernoux has quit [Quit: Leaving]
<d1b2> <TiltMeSenpai> probably want single quotes around that url, there's exclamation points in it. Thanks for the download link though
simukis__ has quit [Quit: No Ping reply in 180 seconds.]
simukis_ has joined #glasgow
simukis_ has quit [Quit: No Ping reply in 180 seconds.]
simukis_ has joined #glasgow
<d1b2> <esden> @whitequark I am downloading it and will put a mirror on my server.
simukis_ has quit [Quit: No Ping reply in 180 seconds.]
simukis_ has joined #glasgow
<whitequark> cool!
<d1b2> <esden> 55% ... wish me luck that it managed to download all 12.51GB
<Ultrasauce> well that's making chromium unhappy in all sorts of interesting ways. thanks for sharing
balrog has quit [Ping timeout: 256 seconds]
<d1b2> <TiltMeSenpai> yeah I think the webapp tries to jam the entire file in ram when you download something
<d1b2> <esden> It is 100% ridiculous....
<d1b2> <esden> I am at 70% 🤞
<d1b2> <esden> they are trying hard to make you install their desktop client I guess?
<d1b2> <gimbas> yep, they do be like that
<hl> I don't recommend trying to download large files using their web client, I've had OOMs from trying that (maybe it works better in chrome though), use megadl
balrog has joined #glasgow
<d1b2> <TiltMeSenpai> oh hi it's done
<sorear> a few years ago I don't think there was a web JS api to stream data other than browser-managed downloads directly to disk
* whitequark shudders
<whitequark> let's... not talk about those APIs